WeatherAlpha 

A weather-driven energy price intelligence engine for ERCOT

🌐 Live Dashboard:https://ashrarxthi.github.io/WeatherAlpha

The Idea

Electricity prices in ERCOT are driven by one core variable: grid tightness, the gap between demand and renewable supply.

• Temperature → drives demand (cooling, heating)

• Wind → drives supply (Texas has ~35GW wind capacity)

• Cloud cover → impacts solar generation

By translating weather → grid tightness → price, WeatherAlpha produces forward-looking signals before they show up in market prices.

How It Works

WeatherAlpha combines three layers:

1. Weather Forecasting
Uses Google DeepMind’s WeatherNext 2 model to generate 15-day forecasts across:

• Temperature

• Wind

• Cloud cover

• Solar conditions

2. Market Calibration
Maps weather conditions to price outcomes using:

• Regression models for price prediction

• Classification models for spike detection

3. Signal Generation
Transforms model output into:

• 15-day price forecasts

• Probability of price spikes

• Structured trading insights powered by AI

What You Get

• 15-day electricity price forecast

• Daily spike probability signals

• AI-generated trading thesis and risk factors

• Interactive dashboard updated automatically

Who This Is For

• Energy hedge funds

• Industrial power buyers

• Insurtech and parametric risk teams

Key Features

• Live ERCOT data ingestion
  Pulls 2 years of day-ahead and real-time market prices

• Advanced weather modeling
  Integrates state-of-the-art forecasting models

• Machine learning calibration
  Learns relationships between weather and pricing

• Spike detection system
  Flags extreme pricing scenarios early

• AI-driven insights
  Converts raw data into actionable strategy

• Auto-updating dashboard
  Fully generated and published on each run

Example Output

Each run generates:

• 15-day forward price curve

• Daily spike probabilities

• Structured signal with:

  • Market thesis

  • Trade ideas

  • Risk considerations

Architecture

The system is built as a modular pipeline:

• Data ingestion (weather + ERCOT pricing)

• Feature engineering (grid tightness modeling)

• ML calibration layer

• Signal generation layer

• Dashboard rendering

Current Status

• ERCOT pricing data: live

• ML model: deployed

• AI signal generation: active

• Dashboard: live

• WeatherNext integration: pending approval

Why This Matters

Energy markets are increasingly driven by real-time environmental dynamics, but most participants rely on lagging indicators.

WeatherAlpha shifts that forward by:

• Modeling the physics behind supply and demand

• Generating signals before price discovery happens

• Turning raw data into actionable insight